A model, utilizing a direct relationship between remotely sensed spectral data and soybean development stage, has been proposed. The model is based upon transforming the spectral data in Landsat bands to greenness values over time and relating the area of this curve to soybean development stage. Analysis of spectral data from soybean research plots revealed that the model works well with reasonable variation in planting date, row spacing, and soil background. Data analyzed from plots whose treatments included two cultivars, two row spacing, and three planting dates had a correlation coefficient (r) of calculated versus observed development stage that exceeded 0.98 for all treatment variables. Likewise, using Landsat-MSS data which contained natural variation due to weather, soil, and other factors, the calculated versus observed development stage gave an r of 0.89 in 1978 and 0.87 in 1979. No difference in the models performance could be detected between early and late planted fields, small and large fields, or high- and low-yielding fields.